In-vivo imaging studies of brain structures have provided valuable information about the nature of neuropsychiatric disorders including neurodegenerative diseases and/or disorders of abnormal neurodevelopment. Driving clinical application of this project is neuroimaging in schizophrenia to study subcortical structures (hippocampus, amygdala, caudate) and ventricular shape. Structural imaging studies have so far most often focused on volumetric assessment of gross brain structures. With increasing evidence for structural changes in small subregions and parts of structures and the avail ability of improved three-dimensional imaging techniques, there is a clear need for providing new image analysis techniques suitable for these tasks. For example, earlier studies of whole ventricular volumes will have to be replaced by subfigural shape studies of parts of ventricles, and there is even evidence that regional shape analysis of the hippocampus structure could differentiate schizophrenics from controls whereas volume differences were not significant. The objective of this project is not specifically to do research on particular hypotheses in regard to these neuroscience problems, but rather to develop and test novel image analysis methodology that would even allow these neuroimage analysis problems to be addressed. First, we will test, validate and apply our new structural image processing methodology based on a multi-level medial model description (m-rep) to substitute manual segmentation. Second, we will derive statistical shape descriptions capturing local and global form in an intuitive, neuroanatomically relevant way. This makes quantitative correlation of shape differences between populations and longitudinal shape changes to intuitive neuroanatomical measurements feasible. Third, we will develop accurate and reliable statistical methods of shape analysis that will provide more powerful discrimination measures of brain pathomorphology than current quantitative morphometric methodologies. Multi-object extension of the m-rep method will be applied to approach the need for composite studies of groups of objects, in particular the hippocampus/amygdala and caudate/ ventricle objects as part of the prefrontal cortical-hippocampal network. The new methods will be applied to subsets of large clinical neuroimaging schizophrenia studies covering prodromal, first episode, treatment responsive and non-responsive and chronic patients. At least 2 of these studies include longitudinal imaging of patients with 3 or 4 time points over a period of 2 years. The increased sensitivity of the new tools has the promise of detecting brain structure differences between different forms of brain disease (neurodevelopmental and neurodegenerative) and (even subtle) changes within patients over the course of their illness. This will reveal a new insight into shape abnormalities of brain structures specific to schizophrenia, via exploring shape variations, anomalies and asymmetries in well-defined patient and control groups.